TOWARDS THE FINE PITCH

Transcription

1 TOWARDS THE FINE PITCH CONSTRAINTS AND METHODOLOGY CROLLES PLANT G PERSICO / D PAPAZIAN / R MILESI STMicroelectronics

2 ST CONTACTS WRITERS FONCTION TEL PART Geraldine PERSICO Probe card Engineer Fine Pitch Evaluation Methodology David PAPAZIAN Product Engineer Contact Resistors Methodology Roger MILESI Maintenance MGR OVERVIEW 2

3 STATUS FINE PITCH AND LOW PAD SIZE INFLUENCE STRONGLY OUR RESULTS IN PRODUCTION IN TERM OF - YIELD (RETEST,BREAK OF PASSIVATION) - USE (EQUIPMENT,IN &OFF SITE CLEANING) - COST (PROBE CARD TIME LIFE,INTERNAL PROCEDURE) TODAY THE CANTILEVER TECHNOLOGY IS FACING COMPROMISE BUT IS STILL ADAPTED TO ALLOW PRODUCTION. - Roger MILESI - 3

4 CONSTRAINTS THE PROBLEMATICS ON CANTILEVER PROBE CARD TECHNOLOGY IS : HOW TO CONTROL THE OVERTRAVEL WHEN THE EFFECTS OF THE ADJUSTMENTS ARE VARIOUS AND OPPOSITE FOR THIS, WE HAVE TO USE A METHODOLOGY WITH PHYSICAL CRITERIAS AND MEASUREMENTS TO FIND OUT AND USE: THE RIGHT LEVEL OF OVERTRAVEL WITH A CORRECT CONTROL OF THE PROCESS (PROBE CARD,PROBER) - Roger MILESI - 4

5 PROBLEMATICS TO SOLVE AT THE WAFER LEVEL PROBING AND PARAMETERS TO SET UP TO HAVE: A GOOD CONTACT RESISTANCE ON WAFER FOR ALL NEEDLES THE SCRUB MUST HAVE A MINIMUM LENGTH SCRUB LENGTH CP MEASUREMENT ON THE SCRUB LENGTH CP ON LINE CONTACT RESISTANCE MEASUREMENT ON THE PRODUCT PARAMETERS: AVERAGE VALUE -> GOOD QUALIY OF THE CONTACT R STANDARD DEVIATION -> STABILITY OF THE CONTACT SIGMA CLEANING -> TO KEEP THE STABILITY - Roger MILESI - 5

6 PROBLEMATICS TO SOLVE AT THE WAFER LEVEL PROBING AND PARAMETERS TO SET UP TO DO NOT HAVE: A CONTACT WITH THE SIDE OF THE PAD SAFE GUARDBAND MUST BE KEEPED TO AVOID ANY BREAK OF PASSIVATION CPK MEASUREMENT ON THE LENGTH AND THE WIDTH OF THE PAD CPK LENGTH CPK WIDTH BREAK OF PASSIVATION - Roger MILESI - 6

7 OVERTRAVEL ADJUSTMENTS RANGE THE OVERTRAVEL IS CONTROLED BY THE PROBER AND BY THE USER. ITS ADJUSTMENT IS LIMITED IN THE LOW RANGE BY A POOR CONTACT RESISTANCE VALUE AND IN THE HIGH RANGE BY A BAD CPK. THE AIM IS TO EXTEND THE ZONE BETWEEN THE TWO CURVES, WITH PROBE CARD USING LIMITED OVERTRAVEL (2 mils standard) and a low overtravel dependence FOR HAVING CONTROLED RESULTS (R measurement, cleaning) IN SPITE OF THE PROBE CARD (planarity) AND PROBER VARIATIONS CONTACT RESISTANCE ON WAFER AREA WITH GOOD CPK CPK 10 OHMS CPK 1.33 AREA WITH GOOD CONTACT RESISTANCE R OVERTRAVEL CORRECT RANGE OF OVERTRAVEL FOR HAVING A GOOD CPK AND A GOOD CONTACT RESISTANCE Standard probe card overtravel: 2 mils - Roger MILESI - 7

8 AIM IS TO HAVE A CONTACT WITH THE LOWEST CONTACT RESISTANCE AND NON DESTRUCTIVE FOR THE PAD PASSIVATION AND FOR THE PROBE 8

9 FINE PITCH PROJECT CHALLENGE SMALLER AND SMALLER PITCH CANTILEVER PROBECARDS LIMIT PROBER ACURACY AND REPEATABILITY SMALLER AND SMALLER PADS FINE PITCH PROCESS CAPABILITY? PRODUCT SENSIBILITY TO CONTACT RESISTORS ANALYSIS METHODOLY SETTING 9

10 VALIDATION METHODOLOGY ambient Temperature Adequate Overtravel adjustment Test on a first TEST CHIP wafer Electrical certification at t0 Mechanical capability at t0 Cp0/CpK0 Bad results Probe card t0 Return to supplier for improvement Contact resistors evolution control Ageing Test on blank wafers: touchdowns Test on a second TEST CHIP wafer high Temperature Probe card Electrical certification after burning Mechanical capability after burning Cp/CpK Bad results Return to supplier for improvement PROBE CARD VALIDATED 10

11 Contact Resistor methodology General Principle To force a current through a needle Current has to be defined by : Tester accuracy, Needles specifications At the moment: I=20mA To measure a voltage on the same needle to extract the mean of the contact resistors Rc values U measured = ( Rc + R others ) * I forced By hypothesis R others is known & steady ( sum of R needle, R pad. Rdyn diode ) At the moment: Rc max= 10 Ohms (clamp value) To extract the 6*SIGMA of the Rc values 11 - David PAPAZIAN -

12 Contact Resistor methodology 3 steps 1- New probe card qualification (on Test Chip) Measure Between two pads linked to a known resistor R contact R Probe-card R ship R contact R Probe-card Force : IDD 20 ma VT measure R contact R Probe-card R wafer 1 R wafer 2 R wafer n..... Force : IDD 20 ma R contact Pad R contactr Probe-card R Probe-card VSS=0V VT measure 3- Production monitoring (on customer product) Vss Pad VSS=0V 2- Ageing step qualification (on blank wafer) Measure contact resistor between 1 needle and all others Measure on a diode pad (like advanced continuity test) VDD=0V VSS=0V Force : IDD 20 ma on Pad R contact R Probe-card VT measure ONLY STEP 1 AND 2 ARE USED IN THE FINE PITCH VALIDATION METHODOLOGY 12 - David PAPAZIAN -

13 PROBECARD OVERTRAVEL ADJUSTMENT Adequate overtravel adjustment on one die: => 6 SIGMA(contact resistors on all pads) < 10 ohms => average contact resistors R stability at this overtravel 14 RCONTACT EVOLUTION VS OVERTRAVEL R CONTACT (OHMS) Stability point 6*SIGMA R overtravel (µm) R(OT) R(OT 5µm) R(OT + 5µm) 13

14 PROBECARD ELECTRICAL CERTIFICATION Probecard Electrical certification on a test wafer => Contact resistors measurement for all probes on all dies: contact resistors database 4. => Standard deviation SIGMA and average value R of each probe contact resistance along the wafer 3.5 Rcontact (Ohms) Probe # Mean R Standard deviation SIGMA electrical validation: 6SIGMA(contact resistors)<10 ohms along the wafer 14

15 CONTACT RESISTORS CONTROL DURING AGEING Adequate overtravel is determined at t0 The same SET UP is conserved during the whole evaluation process If all probe have a contact resistance < CLAMP VALUE: GOOD DIE If at least one probe has a contact resistance > CLAMP VALUE: OPEN PROBLEM Probe does not touch the wafer POSSIBLE SOLUTION REALIGNMENT Probe is too dirty ONLINE CLEANING OFFLINE CLEANING LOW CONTACT RESITORS CAN BE MAINTAINED BY A REGULAR AND ADAPTED ON LINE CLEANING 15

16 PROBECARD MECHANICAL CAPABILITY SCRUB MARKS DATABASE Scrub marks database can be acquired by 2 methods: automatically with the Waferworkx from APPLIED PRECISION or with a manual sampling Number of points Acquisition Advantages Disadvantages WAFERWORKX DATABASE All pads of all dies of the wafer: > points Optic and automatic material Possibility to determine the contribution of process, probe card and prober separately Process capability related to the complete process Need to clean the database to have a better reliability MANUAL DATABASE Sample of 5 dies per wafer and 12 pads per die => 60 points Manual inspection and measurements All datas are checked visually => more reliability Take a long time Scrub marks characteristics: - Scrub length - Vitrif length and vitrif width 16

17 PROBECARD MECHANICAL CAPABILITY Parameters Definition Quality indicators (process capability) in the width and in the length directions of the pad => Calculations of adapted capability coefficients Scrub Length Vitrif length Pad width Vitrif width Pad length * Vitrif = minimum distance between the edge of the scrub and the edge of the pad * Scrub size is defined by the scrub length and the scrub width 17

18 Capability coefficients Definition The Cp represents the width of the distribution between the limits. Cp has to be >2 to have a good homogeneity and reproducibility. The CpK represents the center of the distribution. CpK has to be >1,33 to have a well-centered process. 6 Sigma! For a gaussian distribution Lo Mean Hi Cp = (Hi Lo) / 6 Sigma CpK = Min [ IHi MeanI / 3 Sigma ; IMean LoI/ 3 Sigma) with Sigma = Std deviation 18

19 FINE PITCH CAPABILITY CpK VITRIF LENGTH VITRIF LENGTH MINIMUM VALUE TO HAVE A SECURED PROBING Vitrif length CpK VITRIF LENGTH Capability of the process to have a minimum distance from the scrub to the passivation Vitrif width Pad length Low Limit (µm) 5 High Limit (µm) / The Vitrif width is less critical than the Vitrif Length if no important scrub angle 19

20 FINE PITCH Capability Cp SCRUB LENGTH CONTACT RESISTORS COMPLEXITY OF PROBE CARD LEVELS PLANARITY SCRUB LENGTH VS OVERTRAVEL DEPENDENCE IS RELATED TO SCRUB LENGTH TO HAVE A GOOD CONTACT MINIMUM VALUE LENGTHS HOMOGENEITY Cp SCRUB LENGTH homogeneity of scrub length inside pad limits Low Limit (µm) 5 Average (µm) 25 High Limit (µm) 45 20

21 Capability coefficients EXAMPLES 6 Sigma 6 Sigma Lo Mean GOOD Cp LOW CpK Thin distribution Bad centering Hi Lo Mean GOOD CpK LOW Cp Large distribution Good centering Hi -Cp Vitrif Length correct: good homogeneity of vitrif length values -CpK Vitrif Length very low: vitrif length values under limits -CpK Scrub Length correct: mean value is correct -Cp Scrub Length very low: bad homogeneity of scrub length 21

22 RESULTS AND LIMITATIONS Wafer probe marks analysis have been supported by KUMMER in France, which represents APPLIED PRECISION This methodology has been used with the probe cards suppliers TECHNOPROBE to qualify the WAVE technology (fine pitch technology) and with K&S. ACHIEVEMENTS Objective probecard quality parameters defined: Cp Scrub Length / CpK Vitrif Length Probecard certification methodology Quantified parameters for epoxy probe cards technology improvements Hardware and process control LIMITATIONS FINE PITCH limitation is tightly related to probecard cleaning: ONLINE and OFFLINE 22